Functionalism identifies mental states by their causal-functional roles: what they receive as input, what they produce as output, and how they relate to other mental states. Pain, on this view, doesn’t name a particular neural state; it names the functional role of being caused by tissue damage, producing avoidance behavior, and interacting with beliefs and desires in characteristic ways. Silicon or carbon, neurons or circuits — whatever plays that role is in pain.
Functionalism emerged in the 1960s as a response to the perceived failures of behaviorism (which eliminated inner states) and identity theory (which tied mental states too tightly to particular physical substrates). Hilary Putnam’s multiple realizability argument provided the founding case: if the same mental state can be realized by different physical systems, then mental states cannot be identical to neural state types. This opened the door to computational functionalism, a more specific theory that identifies the relevant functional roles with computational states. Functionalism itself does not say that minds are programs.
The Core Thesis
Putnam’s original formulation (1960s): mental states are multiply realizable. The same mental-state type can be realized by different physical substrates across different organisms and possible systems. A Martian can share our mental states without sharing our neurology, provided its internal organization plays the same functional role.
The functionalist move: define mental states by their input-output profile and their relations to other mental states — not by what they’re made of. That move leaves open how the roles should be specified. Computational functionalism specifies them as states in a machine table or program. The hardware-software picture belongs to that child theory, not to functionalism as such.
What draws philosophers to it:
- Mental states have causal powers. Beliefs and desires don’t just accompany behavior; they produce it in systematic ways.
- Multiple realizability commands wide assent — there is no obvious reason only neurons could think.
- The organizational structure of a system matters for its mental properties, not just the raw physical substrate.
Block’s Objections
Ned Block’s “Troubles with Functionalism” (1978) remains the most influential critique. Two objections define the dialectic:
1. The Absent Qualia Problem
Imagine a system that plays the right functional role (same input-output profile, same internal state transitions) but has no phenomenal experience at all. Block’s homunculus-head: replace each neuron in your brain with a tiny person communicating by two-tin-can telephone. The system as a whole plays the functional role of a human mind. Does it feel anything?
The intuition: probably not. But functionalism cannot say why not — by its own criterion, the homunculus-head qualifies as having mental states.
The book’s diagnosis: the objection bites against narrow, world-bracketed functionalism. An internal diagram can reproduce a pattern of transitions without supplying the history that fixes what its states represent. Even content leaves further questions: which states are conscious, which persisting organization bears them, and what complete world-presenting profile determines the character of states already counted conscious. A system can play a stipulated pain-role without answering any of those questions.
2. The Inverted Qualia Problem
Imagine two people with systematically inverted color experience: what produces “red experience” in you produces “green experience” in your twin, but your behaviors are identical throughout — you both call grass “green” and stop at red lights. Functionally, you’re identical. Phenomenally, you’re inverted.
If functionalism is true, you have the same mental states. But the phenomenal character differs. So phenomenal character outruns functional role.
On the representationalist reading, this argument works against bare internal role as a complete theory of phenomenal consciousness, but it does not establish that qualia are non-physical. Content-fixing history, phenomenal mode and profile, the conscious threshold, and subject individuation require separate specifications rather than one enlarged functional role.
Tye’s Response: Functionalism + Content
Tye’s move (in Ten Problems of Consciousness and “Absent Qualia and the Mind-Body Problem”): functionalism fails not because phenomenal character is non-functional, but because it left out representational content. The right theory combines functional role with what the state represents.
On this account, Block’s stipulated narrow diagram does not establish phenomenal consciousness. A complete assessment must ask what its states represent, whether any enter consciousness, which continuing control organization bears them, and what world-presenting profile conscious states instantiate. A homunculus-head that genuinely satisfied those conditions would qualify; bare transition-role equivalence settles none of them.
Tye’s historical PANIC theory (Poised, Abstract, Nonconceptual, Intentional Content) supplies one influential version of the representational repair. The book retains the identity claim while treating PANIC as one proposal about perceptual form and the conscious threshold, not as a complete account of content or the subject.
The Chinese Room and the Sufficiency Claim
Searle’s Chinese Room targets the computational version of functionalism directly, but its clean result is narrower than Searle’s verdict. The operator’s ignorance shows that rule following by the implementation substrate does not establish operator-level understanding; fluent output does not establish understanding by the implemented whole either. Whether a sufficiently rich virtual organization understands must be judged from its content-fixing relations and coordinated capacities, not settled by the operator’s introspection.
The systems reply correctly relocates the question: perhaps the whole organization understands even though the operator does not. Internalizing the rulebook sharpens the case but does not decide that whole-system attribution. The Room supplies public Chinese content and formal processing; it does not establish that one persisting system integrates those contents across perception, memory, correction, inference, planning, and action. The robot and combination replies sketch how a machine might add the missing organization. The Room therefore defeats syntax-alone sufficiency, not artificial understanding.
My View
Functionalism is the best predecessor to a fully grounded account of mind. It correctly identifies that mental states have causal powers, that material composition alone does not settle mentality, and that organization matters. Computational functionalism adds the proposal that the relevant roles are computational states. More recent work on intrinsic computational functionalism adds a further point worth granting: disciplined causal-computational organization can belong to a physical system independently of an observer’s chosen labels. Narrow functionalism still falls short when it treats an internally specified, world-bracketed causal role as sufficient; its computational child inherits that problem when it treats formal state as sufficient.
| What the view captures | What an internal, world-bracketed description leaves open |
|---|---|
| Functionalism: causal roles and counterfactual organization | What the states represent and what fixes their correctness conditions |
| Functionalism: material neutrality and the possibility of multiple realization | Whether different systems actually realize one mental type |
| Computational functionalism: formal computational structure | Whether formal state determines content or every mind-relevant power |
| Wide functionalism: complex world-involving organization | Whether several capacities support partial understanding, and how broadly and persistently they integrate |
| Material neutrality toward nonbiological realizers | Whether any content has the mode and poising required for consciousness |
My objection therefore targets no fully specified physical-functional account merely because someone calls it functionalist. It targets the internal diagram with the world bracketed. Put the relevant history and producer-consumer relations back in, and the states may acquire genuine content. Let several content-guided capacities constrain one another in an answerable system, and partial understanding can begin. Add breadth, persistence, and traffic across perception, memory, inference, correction, planning, and action, and that understanding becomes more integrated and world-involving. A wide functionalist who includes those relations has reached much the same account by another road; the remaining disagreement concerns filing.
The representational repair remains central: phenomenal character consists in a conscious state’s complete world-presenting intentional profile, not in bare role or mental paint. For artificial understanding, content marks only the first threshold. Content can guide domain-specific competence when it explains flexible inference, correction, learning, planning, or action. Partial understanding begins when several such capacities constrain one another within an answerable system. Greater breadth, depth, persistence, and cross-domain traffic make that understanding more integrated and world-involving; they do not create a separately named grade. Current systems offer credible evidence of domain-specific competence, live but subject-sensitive evidence of partial understanding, and fragmentary evidence of broader integration.
Consciousness and autonomous stake remain separate questions. State consciousness requires an independently supported threshold relation; within a conscious state borne by an identified subject, phenomenal character consists in content under a perceptual or bodily-affective mode, with determinacy, field structure, and attentional organization. Reciprocal self-maintenance may ground autonomy, individuality, practical interest, and welfare. It neither creates content nor constitutes understanding or consciousness. No part of this account privileges neurons, carbon, or biological manufacture.
Related Concepts
- Intentionality — narrow functionalism inherits Brentano’s insight that mental states are directed but lacks a theory of how that directedness gets fixed
- Intentionalism — strong representationalism adds content under the mode, poising, and organization required for phenomenal character
- Qualia — Block’s absent and inverted qualia arguments both target functionalism; my response is representational content, not qualia realism
- The Chinese Room — Searle’s argument defeats syntax-alone sufficiency while leaving whole-system understanding open
- The Turing Test — the test fixes on input-output profile; Block’s lookup-table objection (right behavior, wrong internal organization) is the same lever that troubles functionalism
- Embodiment — bodily engagement supplies one route to world-involving content and perceptual organization, not a biological restriction on minds
- Behaviorism — functionalism arose partly in response to behaviorism’s collapse; it restored inner states while keeping them scientifically tractable
- Identity Theory — functionalism superseded identity theory by introducing multiple realizability; but both leave phenomenal content underspecified
- Semantic Externalism — externalism specifies what functionalism omits: the causal-historical relation between inner states and the world that fixes content
- Teleosemantics — teleosemantics provides one account of how functional states acquire genuine content; a natural complement to functionalism
- Derived and Original Intentionality — content may depend on an interpreting practice or arise from independent world-involving correctness conditions
- Simulation and Realization — disciplined implementation can be intrinsic while meaning, understanding, consciousness, and stake remain separate questions